Urine is an ideal medium in which to focus diagnostic cancer research due to the non-invasive nature and ease of sampling. Many large-scale proteomic studies have shown that urine is unexpectedly complex. We hypothesised that novel diagnostic cancer biomarkers could be discovered using a comparative proteomic analysis of pre-existing data. We assembled a database of 100 published datasets of 5,620 urinary proteins, as well as 46 datasets of 8,620 non-redundant proteins derived from kidney and blood proteome analyses. The data were then used to either subtract or compare molecules from a novel urinary proteome profiling dataset that we generated. We identified 1,161 unique proteins in samples from either cancer-bearing or healthy subjects. Subtractive analysis yielded a subset of 44 proteins that were found uniquely in urine from cancer patients, 30 of which were linked previously to cancer. In conclusion, this approach is useful in discovering novel biomarkers in tissues where unrelated profiling data is available. Only a limited disease-specific novel dataset is required to define new targets or substantiate previous findings. We have shared this discovery platform in the form of our Large Scale Screening Resource database, accessible through the Proteomic Analysis DataBase portal (www.PADB.org).